Clinical studies were performed where all patients underwent esophago gastroduodenoscopy (EGD) by a single endoscopist (SS). Biopsy and spectroscopy sites were measured as distance from the incisor teeth. The optical probe was passed through the accessory channel of the endoscope and placed into light contact with the tissue. Three to four reflectance spectra were collected from each site prior to biopsy. This sequence was repeated in four quadrants at each level at two-centimeter intervals throughout the Barrett's mucosa. An average of 12 sites was evaluated in each patient. Each biopsy was evaluated by an expert pathologist and characterized as containing no dysplasia, low grade dysplasia, high grade dysplasia, indeterminate, or tissue insufficient for diagnosis. All biopsies were also characterized for the presence or absence of inflammatory cells. Each specimen characterized as containing dysplasia or characterized as indeterminate, as well as an equal n umber of non-dysplastic control specimens were submitted to three expert gastrointestinal pathologists for independent and blinded analysis. The collected spectra were analyzed by means of the model and excellent agreement with pathology was found.